Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
External Recruitment as an Incentive
Device
Kong-Ping Chen∗
Institute for Social Sciences and Philosophy
Academia SinicaTaipei, 11529, Taiwan
November 5, 2003
Abstract
External recruitment has often been viewed as a necessary evil in that it trades
off the need for outside talents with the incentives of inside workers. This paper,
however, shows that even from an incentive viewpoint, external recruitment has its
positive role to play. Specifically, if promotion is based on relative performance,
then negative activities in the form of sabotage are a valuable instrument to com-
pete. This results in inefficiency of the workers’ efforts and performance of the firm.
External recruitment, by reducing the marginal return of negative activity relative
to that of productive activity, can restore the incentives of the workers to engage in
productive activity and enhances the firm’s performance. We also show that even
when negative activities are not a concern, external recruitment can sometimes
avoid the shirking equilibrium, or prevents collusion of the workers.
∗I thank Ji-Shyan Wang for initial collaboration, and participants of the 2002 Annual Congress ofEuropean Economic Association in Venice for comments. Financial support from Hong Kong Universityof Science and Technology (DAG\#99/00.BM49) and Taiwan’s National Science Council (89-2415-H-001-042) are gratefully acknowledged.
1. Introduction
Economists have long emphasized the importance of the internal labor market in shielding
workers from external labor market fluctuations.1 The internal labor market is supposed
to function in a way that, except for the designated “port of entry”, open positions
should be filled by internal promotion. This characterization, however, is contradictory
to recent evidence. For example, Baker et al. (1994) show that, within the firm they
study, there is substantial entry in all levels of job.2 Furthermore, the importance of
external recruitment seems to have increased in the past few decades.3 Kanter (1989)
also observes that “climbing career ladder is being replaced by hopping from job to job.”
Given the prevalence of external recruitment and the important role it gradually plays,
it is essential to understand its implied impact on workers’ incentives. An obvious reason
to recruit externally instead of promoting from within is that the outsiders might possess
specific skills or characteristics that the firm needs. (Baron and Kreps, 1999.) That
is, the main reason for a firm to recruit externally is not to provide incentives for the
workers. If anything, it decreases the promotion chance of the inside workers and thus
adversely affects the incentive of insiders to work.4 In other words, it exacerbates the
moral hazard problem within a firm. Even without this moral hazard consideration,
recruited outsiders often lack the firm-specific investment that the insiders have built up
through their careers within the firm. External recruitment is thus a necessary evil in
that it trades off the need for outside talents with insider workers’ incentives.
The purpose of the paper is to show that, contrary to the opinion in the previous1See, for example, the classic treatment in Doeringer and Piore (1985).2They find that over 25 percent of workers entering levels 2-4 (on an 8-level ladder) of the firm are
hired from the outside.3For example, Osterman (1999, p.44) refers to a research project by A. Bernhardt, M. Morris, M.
Handcock and M. Scott reported to the Russell Sage and Rockefeller Foundations showing that there issubstantial increase in job turnover from a cohort of young men entering the labor force in 1966-1981,to another cohort in years 1979-1984. If different cohorts in the sample work for the same span ofyears in lifetime, then there must be increase in external recruitment in the same period. A study ofcareers of managers in telecommunication service by Batt (1996) also documents a greater use of externalrecruitment for middle and upper management positions.
4See, for example, Lazear (1995) and Chan (1996).
2
literature, even if viewed from an incentive perspective, external recruitment also has its
positive function. In most of the firms, promotion is based on relative, rather than ab-
solute, performance.5 In other words, competition for promotion is a tournament in the
sense that, what matters for the workers is not how well they perform, but whether they
outperform others in the same firm.6 Furthermore, in this rank-order tournament, the
winner takes all the prize. Given these natures of promotion tournament, it is not sur-
prising that workers competing for promotion will eventually engage in “sabotages”, the
disruptive behaviors whose main function is only to undercut opponents’ performance.7
The consequence of this behavior is not only that the workers waste resources in unpro-
ductive activities, but also that those who have greatest chance to win the promotion
tournament are not necessarily of the highest caliber.8
By recruiting externally, the firm decreases not only the chance of promotion (and
thus the marginal returns of productive activities as mentioned in previous literature),
but also the effectiveness of negative activities. The latter has two effects on the workers’
incentives. First, as in the case for productive activities, since external competition
reduces the marginal return of negative activities, the level of negative activities will
decrease. The second, and perhaps more important, effect is that while sabotage is almost
useless in competing against outsiders, productive effort remains a useful instrument. The
workers will thus substitute productive for negative effort, which results in an increase in
productive effort and a further reduction of negative activity. We show that although the
total effort of the workers will decrease, this is in fact the net outcome of a reduction of
sabotage and an increase of productive activity. In a word, although external recruitment
hurts the “morale” of insiders by reducing their chance to be promoted, the outputs of5DeVaro (2002) provides evidence showing that it is relative, rather than absolute, performance that
strongly influences chance of promotion.6Green and Stokey (1983), Lazear and Rosen (1981) and Rosen (1986) are seminal works on tourna-
ments. Hvide (2002) and Zabojnik and Bernhardt (2001) are recent theoretical papers, and Bognanno(1999) and Eriksson (1999) recent empirical works.
7See Lazear (1989) for the seminal work on sabotage in a firm, and Chen (2003) for its consequenceon the workers’ incentives and promotion prospects.
8See Chen (2003).
3
the workers actually increase.9
The incentive effect mentioned above is in fact much more general than in the context
when sabotage is a concern. For example, suppose there are two workers, each having a
binary choice of whether to work hard or to shirk. Further assume that no sabotage is
possible in the firm. If there is no external recruitment, then the promotion chance is 1/2
in both the cases when they all work hard and when they all shirk. That is, they suffer
no loss in promotion chance when both shirk. If the disutility of working is substantially
higher than to shirk, then shirking is the unique equilibrium outcome. This, however, is
no longer true with external competition. Since the inside workers are also competing
with outsiders, their promotion probability is lower when they both shirk than when they
both work hard. If the difference in promotion chances is large enough, then the unique
equilibrium becomes the one in which both work hard.
Another possibility is that although it is an equilibrium for the workers to work
hard, they might actually face a Prisoner’s Dilemma in the sense that the utility of both
workers can improve if they can credibly collude to shirk. We will show that even if
this is possible, external recruitment can sometimes reduce their promotion chance by so
much that it is not even profitable to collude. The reason for this is exactly the same:
When they both shirk, no matter by coordination or by collusion, they only give free-
ride to outside competitors. We thus show that external recruitment not only has the
function of reducing sabotage, but also has the power to change the payoff structure of
the promotion tournament game in a way that it breaks shirking as either an equilibrium
or a collusive outcome.10
The remainder of the paper is organized as follows. Section 2 presents the model.9Other works that discuss recruitment from a strategic viewpoint are Chan (1996) and Friebel and
Raith (2001). The first paper shows that external recruitment has negative impacts on the workers’incentives, so an outsider should be given a hurdle, and perform substantially better than insiders inorder to be recruited. The second paper argues that in order to provide middle-level managers incentivesto recruit high quality subordinates, there should be restricted channels of communication in a hierarchy.10A recent paper by Müller and Wärneryd (2001) has shown that introducing outside ownership has
the effect of reducing rent-seeking conflict of members in a production team. The intuition of theirresult is the same as here: External competition reduces the prize of winning, and thus marginal returnrent-seeking behavior.
4
Section 3 discusses the role of external recruitment in shaping the incentives of inside
workers, and shows that it can unconditionally improve the performance of a firm. Section
4 discusses other incentive effects of external recruitment when sabotage is not a concern.
Section 5 concludes.
2. The Model
There are n workers in a firm, and one of them will be promoted to a higher rank. The
criterion for promotion is based on performance, which is a function of the workers’ effort
levels and abilities. Since we are mainly interested in the workers’ incentives in effort, it
is assumed that all workers have the same ability. The performance of a worker increases
with both effort and ability. A simple formulation capturing this idea is to assume that
the performance of worker i is tei + εi, where ei is effort level of worker i, and t is the
ability of the workers. Assume that εi and εj are IID, and f(·) is the density function forεi, i = 1,..., n, which is single-peaked on, and symmetric around, 0 so that E (εi) = 0. εi
can be interpreted as luck of a worker. Our assumption implies that luck is “fair”. Let
Wi = tei be the expected performance of worker i.11 The disutility of effort for worker
i is v(ei), with v0 > 0, v(0) = 0, and v00 > 0. The utility of promotion is u > 0 for
every worker, and is 0 if a worker is not promoted. u can thus be interpreted as wage
gap between ranks. The expected utility of worker i is assumed to be u(ei) = Prob(i is
promoted)u − v(ei),12 and we further assume that the worker with the highest value ofperformance is promoted. In this case we can rewrite u(ei) as
u(ei) = Prob(Wi + εi ≥Wj + εj ∀j)u− v(ei)
= u
Z ∞
−∞Πj 6=iF (Wij + εi)f(εi)dεi − v(ei),
11We will also call Wi the expected output of worker i.12In order to emphasize the rank-order tournament nature of promotion, we have assumed that the
incentive to work is provided solely by the prospect of promotion. In general, there are other ways toreward effort, the most obvious one being the incentive payment where, say, a propotion of output isgiven to the workers. Incorporating these payments will not change the main results of the paper.
5
where F (·) is the distribution function of f(·) and Wij = Wi −Wj is the difference in
expected performance between j and i. Every worker i chooses the value of effort level ei
to maximize his own expected utility. This is a game played among the n workers, and
the symmetric Nash equilibrium is characterized by
(n− 1)utZ ∞
−∞f(ε)2F (ε)n−2dε = v0(e). (1)
It can be easily seen that the equilibrium effort level e∗ rises with the utility of promotion
u and ability t. Let W ∗ be the expected equilibrium performance of a worker, i.e.,
W ∗ ≡ te∗.The workers, however, can also engage in “negative” activities, and “sabotage” other
workers in order to destroy their performance (Lazear, 1989; Chen, 2003). Negative
activities are worthwhile because promotion is based on relative, rather than absolute,
performance. Suppose aij is the level of “sabotage” worker i exerts against j, and that
the expected performance of member i is Wi = tei − g(P
j 6=i aji); where g0 > 0, g00 < 0,
and g(0) = 0. The function g measures the effectiveness of negative effort in destroying
the opponent’s performance. It is a function of the sum of the levels of all other workers’
attack against i, meaning that there is no “synergy” in joining negative efforts. The
disutility of effort for worker i is assumed to be v(ei +P
j 6=i aij), meaning that positive
and negative efforts are “equally undesirable”. Worker i chooses the levels of ei and
aij(j 6= i) to maximize expected utility
u(ei, ai) = Prob (Wi + εi ≥Wj + εj ∀j)u− v(ei +Xj 6=iaij);
where ai ≡ (ai1, ai2, ..., aij−1, aij+1, ..., ain). We will consider the case when both ei andai are strictly positive, so that the symmetric Nash equilibrium is characterized by first-
order conditions
(n− 1)utZ ∞
−∞f(ε)2F (ε)n−2dε = v0(e∗∗ + (n− 1)a∗∗), (2)
ug0((n− 1)a∗∗)Z ∞
−∞f(ε)2F (ε)n−2dε = v0(e∗∗ + (n− 1)a∗∗). (3)
6
We consider only the meaningful case whenW ∗∗ ≡ te∗∗−g((n−1)a∗∗) > 0. From (2) and(3) we know that g0((n− 1)a∗∗) = (n− 1)t. It can be easily seen that a∗∗ decreases withboth t and n. That is, when the ability or number of the workers increases, the level of
every worker’s disruptive activity decreases. Moreover, e∗∗ is increasing in wage gap u.
A direct comparison between (1) and (2) shows that e∗ = e∗∗ + (n − 1)a∗∗. As a result,e∗ > e∗∗, i.e., the workers exert less productive effort, and thus produce lower output
when they can engage in negative activities. Obviously, it follows that W ∗∗ < W ∗. This
result clearly shows that negative activities encroach on the efficiency of a firm: They
divert the efforts of the members from productive to disruptive ones. Not only that, part
of a member’s performance is destroyed by others’ negative effort.
Now suppose that the firm can search for talents from the outside to fill the higher
position. This can be done in two ways. The first is to locate an outsider with proven
performance W . The firm will promote an inside top performer only if his performer
is higher than W ; otherwise this outsider is recruited. Alternatively, the firm can set
a threshold of performance W . An insider is considered for promotion only if his per-
formance is greater than W ; otherwise it commits to search for an outsider to fill the
position. Technically these two practices are equivalent.13 In either case, the expected
utility of worker i is
u(ei, ai) = Pr ob(Wi + εi ≥Wj + εj ∀j, and Wi + εi ≥W )u− v(ei +Xj 6=iaij).
Implicit in this formulation is the assumption that the workers cannot sabotage an out-
sider. It can be justified in several ways. First, the identity of the outsider to be re-
cruited is still unknown at the stage of competition. This is especially true if we adopt
the performance threshold interpretation of external recruitment. In this case it is sim-
ply impossible to attack the potential opponents. Second, even if the identity is known,
because of geographic or informational reasons, it is much more difficult to sabotage an
outsider than an insider. Our qualitative results hold true if instead we assume that it is13For expositional ease we will adopt the second practice in the main body of the paper. The inter-
pretation for Theorem 1, however, are different for the two practices.
7
also possible to attack an outsider, but its effectiveness is sufficiently lower than that for
the insiders. As a result of our assumption, the only way to compete with an outsider is
through higher level of productive effort ei.
Worker i’s chance of promotion can be re-written asR∞W−Wi
Qj 6=i F (Wij + εi)f(εi)dεi,
and the symmetric Nash equilibrium is characterized by
ut{f(W −W )F (W −W )n−1 + (n− 1)Z ∞
W−Wf(ε)2F (ε)n−2dε} = v0(e+ (n− 1)a), (4)
ug0((n− 1)a)Z ∞
W−Wf(ε)2F (ε)n−2dε = v0(e+ (n− 1)a); (5)
where W is the common value of Wi (i = 1, ...n). Denote the solutions of (4) and (5) by
eE and aE, and defineWE = teE−g((n−1)aE) as the equilibrium expected performanceof a worker. Again, for all the discussion in the following we will assume that (4) and (5)
characterize the equilibrium. From (4) and (5) we can show that
ut(n− 1)Z ∞
W−WE
f(ε)2F (ε)n−2dε < ug0((n− 1)aE)Z ∞
W−WE
f(ε)2F (ε)n−2dε.
By the fact that g0((n− 1)a∗∗) = (n− 1)t we have (n− 1)t < g0((n− 1)aE). This impliesthat aE < a∗∗: With external recruitment, the workers will engage in less disruptive
activities. The reason behind this result is quite intuitive. External recruitment increases
the number of workers competing for promotion, and thus reduces the marginal return of
negative activities. The inside workers will therefore respond with less negative activities.
The level of productive effort, however, does not necessarily decrease. As external
recruitment is introduced, there are two forces that influence the level of productive
effort. On the one hand, productive effort is reduced in response to its lower marginal
return. This is exactly the same reason behind the reduction in sabotage. We called
this an “income effect”. On the other hand, since productive effort remains useful in
competing with outsiders while sabotage does not, the extent of reduction in marginal
return is greater for negative than for productive effort. The “substitution effect” will
thus force workers to substitute productive for negative effort. The change in productive
effort will therefore depend on the relative strength of the above two forces.
8
As to its impact on the total effort e+ (n− 1)a, rewrite the left-hand-side of (2) as
(n− 1)utZ W−WE
−∞f(ε)2F (ε)n−2dε+ (n− 1)ut
Z ∞
W−WE
f(ε)2F (ε)n−2dε
= utf(W −WE)F (W −WE)n−1
−utZ W−WE
−∞f 0(ε)F (ε)n−1dε+ (n− 1)ut
Z ∞
W−WE
f(ε)2F (ε)n−2dε. (6)
Comparing (6) with the left-hand-side of (4), we can see that the term −ut RW−WE
−∞
f 0(ε)F (ε)n−1dε is the difference in marginal return of productive effort between the cases
with and without external recruitment. Depending on the value of W and the function
f (·) itself, this term can be either positive or negative. Since f 0(x) > 0 for all x < 0,
we know this term is negative if W 6WE, which in turn implies that eE + (n+ 1)aE <
e∗∗+(n+1)a∗∗ : The total effort is smaller under external recruitment if the performance
thresholdW is chosen to be smaller than the expected total performance of insiders. We
summarize our results in Proposition 1.
Proposition 1 With external recruitment, every worker will engage in less negative
activity. Moreover, when the performance threshold W is not greater than the expected
performance of the inside workers, the workers exert less total effort.
Since promotion chance is at most 1, the expected utility of a worker is bounded
above by u. Moreover, since both aE and eE can take only non-negative values and since
v00 > 0, aE and eE must also be bounded. W −WE thus approaches infinity, and the
left-hand-side of (5) approaches 0, when W → ∞. Therefore, there exists W such that
aE = 0 for W > W . That is, if the value of W is high enough so that promotion is
sufficiently difficult, then sabotage disappears. This, however, does not imply that the
firm should optimally set a performance threshold high enough to eradicate sabotage.
The reason is that a high performance threshold also implies low productive effort e. As
a result, it might not be optimal for a firm to choose a W that can eliminate sabotage,
because it might at the same time induce very low productive effort. In the next section
we set out to find the optimal value of W that maximizes the firm’s output.
9
3. The Optimal External Recruitment Policy
In this section we investigate the role of external competition in shaping the workers’
incentives in (both productive and negative) efforts. We then derive the optimal perfor-
mance threshold (W ) and wage gap (u) of the firm and the optimal recruitment and wage
policies they imply. To emphasize the dependence of aE, eE andWE on the performance
threshold W , we will write them as aE(W ), eE(W ) and WE(W ).
The performance threshold, W , obviously has a great bearing on the incentives of
the workers, and eventually on the output of the firm. For example, as W → −∞,every insider makes the threshold for sure, and the case is equivalent to when the firm
only promotes internally.14 That is, limW→−∞
eE(W ) = e∗∗, limW→−∞
aE(W ) = a∗∗ and
limW→−∞
WE(W ) = W ∗∗. When W → ∞, the promotion chance of inside workers goesto zero, and the optimal effort levels are thus corner solutions with eE = aE = 0. Con-
sequently, limW→∞
WE(W ) = 0. We first investigate the effects of W on the workers’ effort
levels.
Proposition 2 ∂e/∂W > 0 and ∂a/∂W < 0 if W 6 WE(W ).
Proof: See the Appendix.
Proposition 2 says that a worker’s productive (negative) effort increases (decreases)
with the level of performance threshold, if the value of the threshold is not higher than the
expected performance of insiders. Since the case when there is no external competition
corresponds to when W = −∞, which in turn implies effort level e∗∗and a∗∗, if the firmgradually raises the performance thresholdW from a very low value, then by Proposition
2 a worker’s negative effort will gradually decrease from a∗∗ , and productive effort
increases from e∗∗. In other words, as long as W 6 WE, the effort levels will be such
that eE(W ) > e∗∗ and aE(W ) < a∗∗. Since WE(−∞) = W ∗∗ > 0, we know that
WE(W ) > W ∗∗, and is increasing in W , as long as W 6WE(W ). This in turn implies
that the value of W that maximizes output WE must be such that WE(W ) > W ∗∗. The
reason for this is quite simple: since W ∗∗ is independent of W , andWE(W ) is increasing14This can be easily seen by the fact that when W →−∞, (4) reduces to (2) and (5) to (3).
10
when W 6 WE(W ), the value of W that maximizes WE(W ) cannot possibly be below
WE(W ), as the firm can still raise W to increase output. We thus have:
Proposition 3 The value of the performance threshold that maximizes WE(W ), Wo, is
such that WE(Wo) > W ∗∗ and WE(W
o) < W
o.
Proof: See Appendix.
The function WE(W ) can thus be depicted as in Figure 1.
The intuition for Proposition 3 is as follows. When the performance threshold is
raised, the marginal returns for both productive and negative effort decreases. We call
the corresponding reduction in efforts the “income effect”. However, there is another force
that affects the value of efforts. While sabotage is not useful against outsiders, productive
effort remains an useful instrument in competing with the outsiders. The workers will
thus substitute productive for negative effort. While this “substitution effect” might not
dominate the income effect, it ensures that the reduction in (output-destroying) sabotage
is large enough to increase the total output.
Proposition 3 implies that, if we adopt the first interpretation of external recruitment
mentioned in Section 2, then outsiders should be given a “hurdle” in competing with
inside workers. That is, the firm only considers outsiders whose performance is higher
than that expected of the insiders as potential candidates to be recruited. This is con-
sistent with Chan (1996), which proves the same result in the context without sabotage.
There are, however, two differences that need to be noted. First, while in Chan (1996)
productive effort necessarily decreases when there is external recruitment, in our model
it is not necessarily so. In fact, not only productive effort, but also total effort might very
well increase. Second, in Chan (1996) external recruitment is an evil that the firm must
bear (productive effort decreases), while in our model it actually increases total output
at the optimum.
If we adopt the second interpretation of external recruitment, then since Wo>
WE(Wo), the inside workers must perform better than expected to be promoted. That
is, only ones who perform exceptionally well are to be considered as candidates for pro-
11
-
6
WE
W ∗∗
WW
o
45◦
W 0
WE(W )
¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡
Figure 1 Output as a function of performance threshold
12
motion. Note that since we assume that all workers have the same ability, they actually
have the same expected performances. As a result, only those who are lucky will make
the threshold Wo. In this sense external recruitment is purely used to lessen incentives
for sabotage. Put differently, it is purely an incentive device.
Proposition 3 is of great significance in that it fully justifies external recruitment on
incentive grounds, for external recruitment ameliorates the problem of infight without
reducing incentives toward productive activities. It should be emphasized that in contrast
to the traditional literature, in our model external recruitment increases, rather than
reduces the productive effort.
It should also be emphasized, however, that despite the result in Proposition 3, the
objective of the firm is not to maximize output. This is because in order to increase
output, it might be that the worker’s total effort, e+ (n− 1)a, has to be increased. (Wewill show shortly that as W increases, total effort e+ (n− 1)a will indeed increase.) Inthat case the disutility of effort increases, and the workers need to be paid more in order
to raise output. In particular, if the worker is paid a reservation utility that equals the
total disutility of effort, then the increase in wage payment might outweigh the increase
in output thus produced. In other words, the firm might pay too high a price to maximize
output WE. Following Lazear and Rosen (1981) and Chan (1996), we will thus assume
that the objective of the firm is to maximize the sum of workers’ utilities subject to
zero-profit constraint:
maxW,u
nXi=1
pEi u−nXi=1
vEi
s.t. u =nXi=1
WEi (W );
where pEi is the promotion probability of worker i, and vEi ≡ v(eEi +(n−1)aEi ). Substitut-
ing the zero-profit constraint into the objective function, we know that in the symmetric
equilibrium the firm maximizes
13
nXi=1
pEi
nXi=1
WEi (W )−
nXi=1
vEi
= n(npEWE(W )− vE) ≡ nπ(W,u).
The firm thus chooses W and u in order to maximize π(W,u). Denote the optimal
values of W and u by WEand uE, respectively. The equilibrium promotion probability
of a worker is thus pE(WE) = (1−Fn(WE−WE(W
E)))/n < 1/n. The following lemma
proves useful:
Lemma 1 ∂(eE + (n− 1)aE)/∂W > (<) 0 if W < (>) WE(W ).
Proof: See Appendix.
With lemma 1 we can now compare the optimal outputWE(WE) with the equilibrium
output without external recruitment, W ∗∗. Suppose thatWE(WE) 6 W ∗∗. From Figure
1 we can see clearly that this is true only ifWE >W
0, whereW
0is such thatWE(W
0) =
W ∗∗. In that case WE > W
0> W
o. From Proposition 3 and Lemma 1 we know that
eE+(n−1)aE > e∗∗+(n−1)a∗∗, which in turn implies that vE > v∗∗ ≡ v(e∗∗+(n−1)a∗∗).It follows that π(W,u) = npEWE(W
E) − vE < WE(W
E) − vE 6 W ∗∗ − v∗∗. This
contradicts the definition of WE, which is chosen to maximizeWE−vE.15 We thus must
have WE(WE) > W ∗∗. That is, the output level under the optimal external recruitment
policy is greater than that without external recruitment.
It is worthwhile to note that while external recruitment can improve the performance
of the firm, it is not strong enough to totally eliminate the harm caused by negative
activities. This can be easily seen by comparing (4) and (2), which show that e∗ >
eE(WE) + (n− 1)aE (WE
). As a result, W ∗ > WE(WE). We thus have
Proposition 4 W ∗ > WE(WE) > W ∗∗. That is, external recruitment can ameliorate,
but cannot eliminate, the harm caused by sabotage.
Note that zero-profit condition implies that uE = WE(WE)/n. Similarly, zero-profit
conditions for the cases with and without sabotage are u∗ = W ∗/n and u∗∗ = W ∗∗/n,15Note that W ∗∗ =WE(−∞) and v∗∗ = lim
W→−∞vE.
14
respectively. Since WE(WE) > W ∗∗, we know that uE > u∗∗. This means that in
the presence of sabotage, the optimal wage gap is greater when the firm can recruit
outsiders. The reason for this is straightforward. Since external recruitment can mitigate
the problem of infight, the firm can afford to design a higher-powered incentive contract
by setting a wider wage gap in order to induce higher productive effort. A simple corollary
of Proposition 4 is thus:
Proposition 5 In case the firm makes zero profit, it follows that u∗ > uE > u∗∗.
The result that u∗ > u∗∗ is exactly the main point made in Lazear (1989), which
argues that in order to mitigate sabotage within a firm, it should narrow the wage gaps
between hierarchies, despite the fact that it might simultaneously weaken their incentives
to work.
The first order condition for u under the case without external recruitment is
∂W ∗∗
∂u= (
∂e∗∗
∂u+ (n− 1)∂a
∗∗
∂u)v0(·).
The optimal choice of u balances the trade-off between the benefit (left-hand side) of in-
creasing u and its cost (the right-hand side). On the other hand, the first-order condition
under the case with external recruitment is
∂npE
∂uWE + npE
∂WE
∂u= (
∂eE
∂u+ (n− 1)∂a
E
∂u)v0(·).
The marginal cost of increasing the value of u (right-hand side) is the same as in the case
without external recruitment. There are, however, two sources of benefit in increasing
u now. As in the case without external recruitment, it influences the effort levels, and
thus outputs, of the workers (the second term on the left-hand side). Moreover, since
the change in effort will affect absolute output levels, it also changes the promotion
chance of the inside workers (the first term on the left-hand side), as it changes the
output level relative to outsiders. This is the term that is absent in the case without
external recruitment, as the promotion probability is always 1/n regardless of the value
of u. Since e increases the absolute value of output while a decreases it, the only way to
increase the probability of outperforming W is through the increase and e (and perhaps
15
reduction in a). Thus external recruitment shifts the workers’ efforts from sabotage back
to productive activities by forcing them to take into consideration the relative benefit of
productive and negative effort in influencing their chance of promotion.
4. Other Incentive Effects
So far our discussion is focused on the function of external recruiting in reducing disrup-
tive behavior in an organization. However, sometimes external recruitment can improve
incentives even if negative activity is not a concern. This is especially so when the effort
choice of the workers is discrete.
Suppose there are two workers in the firm vying for promotion. The workers can
either exert high (H) or low (L) effort level. Assume that the output of the firm when
they both exert high effort is so high that the firm prefers (H,H) to (L,L), after paying
for the wages to the workers. As a result, the firm would like (H,H) to be implemented.
Also assume that the workers do not engage in negative activities. The disutility of high
(low) effort level is x (y), where x > y > 0. Let pij (i, j = H,L) be the probability
that a worker is promoted when his effort level is i and that of his opponent is j. The
utility of being promoted is u, and is 0 otherwise. The expected utility of worker i is thus
ui(i, j;u) = piju− x if i = H and is piju− y if i = L. The normal form expression of thegame between the two workers is given in Table 1, where player 1 (2) is row (column)
player.
H L
H pHHu− x, pHHu− x pHLu− x, pLHu− yL pLHu− y, pHLu− x pLLu− y, pLLu− y
Table 1 Payoff matrix of workers
The key insight is that without external recruitment, each of the workers has a pro-
motion chance of 1/2 as long as they exert the same effort level, regardless of high or low.
16
(Specifically, without outside competition for promotion, the promotion chance of each
worker is 1/2 under both (H,H) and (L,L).) If the disutility of effort is substantial, then
(L,L) will be a Nash equilibrium. However, when there is outside competition, shirk-
ing of insiders will give the outsider a greater chance of being recruited. That means
although external recruitment reduces the promotion chance of workers when they exert
high effort levels, the decrement is not as large as when both shirk. If the reduction in
promotion chance is substantial, then external recruitment shifts the Nash equilibrium
of the game from (L,L) to (H,H).16
To give a concrete example, assume that pHH = 1/2, pHL = 11/20, pLH = 9/20, and
pLL = 1/2. It is easy to see that as long as x − y > u/20, then (L,L) is the unique
Nash equilibrium.17 Suppose by introducing external competition the firm can change
promotion probabilities of the workers to pHH = 1/3, pHL = 1/4, pLH = 1/12, and
pLL = 1/10. Then (H,H) will be the only equilibrium if 320u > x − y.18 Numerically,
if u = 40, x = 10 and y = 5, then (L,L) is the unique equilibrium when there is
only internal promotion; and (H,H) is the unique equilibrium when there is external
recruitment. Thus external recruitment can change the payoff structure of the game so
that the equilibrium shifts from the inefficient (L,L) to the efficient one, (H,H). The
main specifications that drive this result are that (a) without external recruitment, the
promotion chance of each worker is 1/2 under either (H,H) or (L,L); and (b) with
external recruitment, the promotion chance of each worker decreases, but the drop in
promotion probability is greater for (L,L) than (H,H), as 1/2− 1/10 > 1/2− 1/3.Another function of external recruitment is concerned with collusion. As is suggested
by Lazear (1995), promotion based on relative performance results in not only non-
cooperation and mutual sabotage, but also harmful cooperative behavior. In some cases,
even if the firm designs the prize of promotion in such a way that exerting high level of16Note that this result is impossible when the effort level is continuous. As can been seen from a
slightly modified equation (1) that allows for external recruitment, the effort level always decreaseswhen there is external recruitment. This means that our results in this section crucially depend on thediscreteness assumption of effort levels. However, in reality this might not be an uncommon case.17In fact exerting effort level L is the dominant strategy for every worker.18In this case, exerting effort level H is the dominant strategy for every worker.
17
effort is actually a Nash equilibrium, there is still possibility that the workers can collude
to shirk and enjoy the gain from collusion.19 For example, in discussing the compensation
practice of the banking-wiring room in the Hawthorne plant of Western Electric, Miller
(1992) describes a game called “binging” played between workers that is actually used
to punish the workers who produces too much; that is, it is a collusive device to prevents
the workers from exerting high effort level.20 In this case, external recruitment is effective
in reducing incentive to collude, because the presence of external competitor will reduce
the promotion chance of all the inside workers if they collude to shirk. A simple example
to explain this is as follows.
In Table 1, let PHH = PLL = 1/2, u = 24, x = 4, and y = 2. Then there is a
unique Nash equilibria of the game, (H,H), where each receives a utility of 8. However,
although (L,L) is not a Nash equilibrium, each worker can receive a higher utility of
10 under that strategy profile. Under this specification, Table 1 is actually a game of
Prisoners’ Dilemma, in which players are both better off when they cooperate to shirk,
but non-cooperation (in which they both work hard) is the unique equilibrium. The
workers thus have incentives to collude and coordinate on (L,L). In our context, this
can be achieved with the following side-contract: The workers agree that whoever wins
the contest pays the loser an amount which equals half of the prize of winning, 24. In
this case the workers’ utilities is independent of the outcome of the tournament. Since
payoff is now independent of outcomes, they all exert low effort level L. The utility
of each is thus 1/2 × 24 − y = 12 − 2 = 10, which is exactly the collusive outcome.
Note that this side-contract is verifiable and legal, and is thus enforceable.21 Again,
we assume that the output of the firm when every worker exerts high effort level is so19This is exactly the same kind of collusive behavior that occurs in competitive bidding.20See also the discussion in Gibbons and Waldman (1999), Section IV.B.21In a more complicated context, the enforcement of collusive behavior (and, more generally, side-
contract) might be a difficult problem. This area of study is still on the frontier of research with noconsensus yet (See, for example, Tirole, 1992 and Felli and Villas-Boas, 1998). For our purpose, an easyway to sustain (L,L) as a collusive result is to imbed the model in a repeated game context. In thatcase, a trigger strategy in the “relational contract” (Bull, 1987; Levine, 1993) can be used to sustain thecollusive outcome among players.
18
desirable that it wants to implement (H,H). Thus collusion between the workers is
harmful from the firm’s perspective. By introducing external competition the firm can
sometimes eliminate incentives of the workers to collude. For example, suppose that
with external recruitment, promotion probability of inside workers is 0 when they all
exert low level of effort, and is only slightly lower than 1/2 when they exert high levels of
effort. Then not only is (H,H) still the unique equilibrium, but also the workers now do
not have incentives to collude towards (L,L), because the utility of each worker under
this outcome is negative, and is thus Pareto-dominated by (H,H). The reason for this
result is, again, that shirking is more costly for the workers in the presence of external
competition than without. When only insiders compete for promotion, reducing effort
will not decrease the promotion probability of a worker, as long as all the other workers
follow suit. When there is outside competition, even if every inside worker can collude to
shirk, it only treats the outsider to a free lunch. Sometimes the loss of promotion chance
is so large that the workers have no choice but to exert high effort level. As a result,
external recruitment recovers (H,H) not only as the unique Nash equilibrium, but also
a Pareto-dominant outcome which is immune to collusion.
5. Conclusion
In this paper we show that, contrary to the common belief that external recruitment
trades off the need for the outside talent with incentives of inside workers, external
recruitment can unconditionally improve the performance of a firm. This is because
external excruitment, by reducing the marginal return of sabotage by more than that
of productive activities, can force the workers to substitute the former for the latter.
As a result, the output of the firm increases. We go on to show that, even if sabotage
is not a concern of the firm, external recruitment can still be a valuable practice in
recovering incentives. For a firm stuck in low-effort equilibrium, the introduction of
external competition might be able to force the workers to pay a higher price for shirking,
by so much so that it shifts the low effort equilibrium to a high effort one. For example, in
19
an academic department, the junior faculty will probably work harder (in either teaching
or research) when the department practices senior recruiting, than when they are sure
that a senior position will be filled by one among them.
In some cases the game of promotion tournament might be characterized by a Pris-
oner’s Dilemma, in which the equilibrium (with high effort level) is Pareto-dominated
by a non-equilibrium outcome (with low effort level) for the players. In that case the
workers will have incentives to collude and coordinate on the latter outcome, which gives
them higher utility. The firm can break this collusion by introducing external competi-
tion. The reason for this is exactly the same: without external competition, the workers
pay no price when they both shirk. When there is external recruitment, they are giving
free-rides to outsiders if they collude to shirk.
By investigating the incentive aspect of external recruitment, this paper adds to the
so far relatively scant literature regarding strategic approach to recruitment practice.
As future research, it greatly enhances our understanding of the tournament aspect of
promotion if we can combine the incentive and the traditional adverse selection consid-
erations of external recruitment in an integrated model.
20
Appendix
(i) Proof of Proposition 2:
Differentiating (4) and (5) with respect to W, and solving for ∂e/∂W and ∂a/∂W we
have22
∂e
∂W= (n− 1)utf
0Fn−1[v00 − ug00 R∞W−WE f
2(ε)F (ε)n−2dε] + ug0f 2F n−2v00
∆, (A1)
∂a
∂W=u F n−2v00{g0f 2 + tf 0F}
−∆ ; (A2)
where ∆ = −(n − 1)[ut2f 0Fn−1 + v00][ug00 R∞W−WE f
2(ε)F (ε)n−2dε + ug02f 2F n−2 − v00] +(n−1)[utg0f 0F n−1+v00][utg0f 2Fn−2−v00], which is positive by the second-order condition.Since f 0(x) > 0 if and only if x < 0, we know from (A1) and (A2) that ∂e/∂W > 0 and
∂a/∂W < 0 if W < WE(W ).
(ii) Proof of Proposition 3:
The first order condition for WEis
∂WE
∂W= t
∂eE
∂W− (n− 1)g0∂a
E
∂W
= (n− 1)∆−1 (tv00 − tug00 R∞
W−WE f(ε)2F (ε)n−2dε+ g0v00)utF n−1f 0
+(t+ g0)uFn−2v00f2g0
= 0.
The second term in the brace is positive, and the term in the parenthesis of the first term
is also positive. Consequently, f0(W
E −WE(WE)) must be negative. Since f 0(x) > 0 if
and only if x < 0, this implies that WE(WE) > W
E.
(iii) Proof of Lemma 1:
From (A1) and (A2) we know that
∂(eE + (n− 1)aE)∂W
= −u2tF n−2f
0g00 R∞
W−WE f2(ε)F (ε)n−2dε
∆.
22For brevity we have omitted the variables in functions v(e), F (W −WE), and f(W −WE).
21
Obviously, the expression above is positive (negative) if f0(W −WE(W )) > (<) 0, which
in turns implies W < (>) WE(W ).
22
References
[1] Baker, G., M. Gibbons and B. Holmstrom (1994), “The Internal Economics of the
Firm: Evidence from Personnel Data, Quarterly Journal of Economics, 109, 881-919.
[2] Baron, J. N. and D. M. Kreps (1999), Strategic Human Resources: Frameworks for
General Managers, John Wiley, New York, NY.
[3] Batt, R. (1996), “From Bureaucracy to Enterprise? The Changing Jobs and Careers
of Managers in the Telecommunication Service,” in Broken Ladder: Managerial
Careers in the New Economy, P. Osterman ed., Oxford University Press, New York,
NY.
[4] Bognanno, M. L. (1999), “Corporate Tournament,” Journal of Labor Economics, 19,
290-315.
[5] Bull, C. (1987), “The Existence of Self-Enforcing Implicit Contracts,” Quarterly
Journal of Economics, 147-60.
[6] Chan, W. (1996), “External Recruitment versus Internal Promotion” Journal of
Labor Economics, 14, 555-70.
[7] Chen, K. -P. (2003), “Sabotage in Promotion Tournaments,” Journal of Law, Eco-
nomics and Organization, 19, 119-40.
[8] Doeringer, P. and M. Piore (1985), Internal Labor Markets and Manpower Analysis,
M. E. Sharpe, UK.
[9] DeVaro, J. (2002), “Promotion Tournament in Real Firms,” mimeo, Cornell Univer-
sity.
[10] Eriksson, T. (1999), “Executive Compensation and Tournament Theory: Empirical
Test on Danish Data,” Journal of Labor Economics, 17, 262-80.
23
[11] Felli, L. and M. Villas-Boas (2000), “Renegotiation and Collusion in Organization,”
Journal of Economics and Management Strategy, 9, 453-83..
[12] Friebel, G. and M. Raith (2001), “Abuse of Authority and Hierarchical Communi-
cation,” forthcoming in RAND Journal of Economics.
[13] Gibbons, R. and M. Waldman (1999), “Careers in Organizations: Theory and Ev-
idence,” in O. Ashenfelter and D. Card ed., Handbook of Labor Economics Vol. 3,
North Holland, Netherlands.
[14] Green J. and N. Stokey (1983), “A comparison of Tournaments and Contests,”
Journal of Political Economy, 91, 349-64.
[15] Hvide, H. K. (2002), “Tournament and Risk Taking,”, Journal of Labor Economics,
20, 877-98.
[16] Kanter, R. (1989),When Giants Learn to Dance, Simon & Schuster, New York, NY.
[17] Lazear, E. P. (1989), “Pay Equality and Industrial Politics,” Journal of Political
Economy, 97, 561-80.
[18] ––– (1995), Personnel Economics, MIT Press, Cambridge, MA.
[19] ––– and S. Rosen. (1981), “Rank-Order Tournament as Optimum Labor Con-
tract,” Journal of Political Economy, 89, 841-64.
[20] Levin, J. (2003), “Relational Incentive Contracts,” American Economic Review, 93,
835-57.
[21] Miller, G. (1992), Managerial Dilemma: The Political Economy of Hierarchy, Cam-
bridge University Press, New York, NY.
[22] Müller, H. and K. Wärneryd (2001), “Inside versus Outside Ownership: A Political
Theory of the Firm,” RAND Journal of Economics, 32, 527-41.
24
[23] Osterman, P. (1999), Securing Prosperity: The American Labor Market: How it has
Changed and What to do About it, Princeton University Press, Princeton, NJ.
[24] Rosen, S. (1986), “Prizes and Incentives in Elimination of Tournaments,”American
Economic Review, 76, 701-15.
[25] Tirole, J. (1992), “Collusion and the Theory of Organizations,” in J.-J. Laffont ed.,
Advances in Economic Theory Vol.2, Cambridge University Press, Cambridge, UK.
[26] Zabojnik, J. and D. Bernhardt (2001), “Corporate Tournaments, Human Capital
Acquisition, and the Firm Size-Wage Relation,” Review of Economic Studies, 68,
693-716.
25